Herbimycin A and geldanamycin (HA and GA, respectively) are benzoquinoid ansamycin antibiotics which have been reported to reverse the transformed phenotype of virus-transformed epithelial cells. While these agents are weak inhibitors of tyrosine kinases in vitro, a result of their addition to cells is delayed reduction in cellular protein phosphotyrosine content and in the activity of certain tyrosine kinases. The focus of this project is to examine these agents' cytotoxicity against cell lines derived from the neuroectoderm and against those cell lines with neurocrine features. Secondarily, our focus is to determine whether any observed cytotoxicity is due to tyrosine kinase inhibition or to another, as yet undefined, mechanism. We have determined that both drugs are very cytotoxic against both primitive, undifferentiated neuroectoderm-derived tumors and tumors with neurocrine properties. These tumors include medulloblastomas, neuroepitheliomas, colon carcinomas, melanomas, and prostatic carcinomas. More differentiated neuroectoderm-derived cells such as neuroblastoma, are not affected. Many other cell lines, representative of the hematopoietic system and fibroblasts, are not sensitive at the concen- trations employed. We have been able to demonstrate in vivo effectiveness of these drugs as well. Using a subcutaneous athymic mouse-human xenograft model, we observed a significant reduction in tumor mass following subcutaneous drug administration. Tumors studied were prostatic carcinoma (hormone-refractory) and neuroepithelioma. No overt toxicity to the host was seen. We have also demonstrated that these drugs are cytotoxic to a primary explant of a human neuroepithelioma, while not affecting the viability of normal rat brain cortical and cerebellar cultures. The toxicities observed in all cases require an exposure of cells to drug of as little as one hour.